src/Protocol/PortableContact.php

   1 <?php
   2 /**
   3  * @file src/Protocol/PortableContact.php
   4  *
   5  * @todo Move GNU Social URL schemata (http://server.tld/user/number) to http://server.tld/username
   6  * @todo Fetch profile data from profile page for Redmatrix users
   7  * @todo Detect if it is a forum
   8  */
   9
  10 namespace Friendica\Protocol;
  11
  12 use DOMDocument;
  13 use DOMXPath;
  14 use Exception;
  15 use Friendica\Content\Text\HTML;
  16 use Friendica\Core\Config;
  17 use Friendica\Core\Protocol;
  18 use Friendica\Core\Worker;
  19 use Friendica\Database\DBA;
  20 use Friendica\Model\GContact;
  21 use Friendica\Model\Profile;
  22 use Friendica\Network\Probe;
  23 use Friendica\Util\DateTimeFormat;
  24 use Friendica\Util\Network;
  25 use Friendica\Util\XML;
  26
  27 require_once 'include/dba.php';
  28
  29 class PortableContact
  30 {
  31         /**
  32          * @brief Fetch POCO data
  33          *
  34          * @param integer $cid  Contact ID
  35          * @param integer $uid  User ID
  36          * @param integer $zcid Global Contact ID
  37          * @param integer $url  POCO address that should be polled
  38          *
  39          * Given a contact-id (minimum), load the PortableContacts friend list for that contact,
  40          * and add the entries to the gcontact (Global Contact) table, or update existing entries
  41          * if anything (name or photo) has changed.
  42          * We use normalised urls for comparison which ignore http vs https and www.domain vs domain
  43          *
  44          * Once the global contact is stored add (if necessary) the contact linkage which associates
  45          * the given uid, cid to the global contact entry. There can be many uid/cid combinations
  46          * pointing to the same global contact id.
  47          *
  48          */
  49         public static function loadWorker($cid, $uid = 0, $zcid = 0, $url = null)
  50         {
  51                 // Call the function "load" via the worker
  52                 Worker::add(PRIORITY_LOW, "DiscoverPoCo", "load", (int)$cid, (int)$uid, (int)$zcid, $url);
  53         }
  54
  55         /**
  56          * @brief Fetch POCO data from the worker
  57          *
  58          * @param integer $cid  Contact ID
  59          * @param integer $uid  User ID
  60          * @param integer $zcid Global Contact ID
  61          * @param integer $url  POCO address that should be polled
  62          *
  63          */
  64         public static function load($cid, $uid, $zcid, $url)
  65         {
  66                 $a = get_app();
  67
  68                 if ($cid) {
  69                         if (!$url || !$uid) {
  70                                 $contact = DBA::selectFirst('contact', ['poco', 'uid'], ['id' => $cid]);
  71                                 if (DBA::isResult($contact)) {
  72                                         $url = $contact['poco'];
  73                                         $uid = $contact['uid'];
  74                                 }
  75                         }
  76                         if (!$uid) {
  77                                 return;
  78                         }
  79                 }
  80
  81                 if (!$url) {
  82                         return;
  83                 }
  84
  85                 $url = $url . (($uid) ? '/@me/@all?fields=displayName,urls,photos,updated,network,aboutMe,currentLocation,tags,gender,contactType,generation' : '?fields=displayName,urls,photos,updated,network,aboutMe,currentLocation,tags,gender,contactType,generation') ;
  86
  87                 logger('load: ' . $url, LOGGER_DEBUG);
  88
  89                 $s = Network::fetchUrl($url);
  90
  91                 logger('load: returns ' . $s, LOGGER_DATA);
  92
  93                 logger('load: return code: ' . $a->get_curl_code(), LOGGER_DEBUG);
  94
  95                 if (($a->get_curl_code() > 299) || (! $s)) {
  96                         return;
  97                 }
  98
  99                 $j = json_decode($s, true);
 100
 101                 logger('load: json: ' . print_r($j, true), LOGGER_DATA);
 102
 103                 if (!isset($j['entry'])) {
 104                         return;
 105                 }
 106
 107                 $total = 0;
 108                 foreach ($j['entry'] as $entry) {
 109                         $total ++;
 110                         $profile_url = '';
 111                         $profile_photo = '';
 112                         $connect_url = '';
 113                         $name = '';
 114                         $network = '';
 115                         $updated = NULL_DATE;
 116                         $location = '';
 117                         $about = '';
 118                         $keywords = '';
 119                         $gender = '';
 120                         $contact_type = -1;
 121                         $generation = 0;
 122
 123                         if (!empty($entry['displayName'])) {
 124                                 $name = $entry['displayName'];
 125                         }
 126
 127                         if (isset($entry['urls'])) {
 128                                 foreach ($entry['urls'] as $url) {
 129                                         if ($url['type'] == 'profile') {
 130                                                 $profile_url = $url['value'];
 131                                                 continue;
 132                                         }
 133                                         if ($url['type'] == 'webfinger') {
 134                                                 $connect_url = str_replace('acct:', '', $url['value']);
 135                                                 continue;
 136                                         }
 137                                 }
 138                         }
 139                         if (isset($entry['photos'])) {
 140                                 foreach ($entry['photos'] as $photo) {
 141                                         if ($photo['type'] == 'profile') {
 142                                                 $profile_photo = $photo['value'];
 143                                                 continue;
 144                                         }
 145                                 }
 146                         }
 147
 148                         if (isset($entry['updated'])) {
 149                                 $updated = date(DateTimeFormat::MYSQL, strtotime($entry['updated']));
 150                         }
 151
 152                         if (isset($entry['network'])) {
 153                                 $network = $entry['network'];
 154                         }
 155
 156                         if (isset($entry['currentLocation'])) {
 157                                 $location = $entry['currentLocation'];
 158                         }
 159
 160                         if (isset($entry['aboutMe'])) {
 161                                 $about = HTML::toBBCode($entry['aboutMe']);
 162                         }
 163
 164                         if (isset($entry['gender'])) {
 165                                 $gender = $entry['gender'];
 166                         }
 167
 168                         if (isset($entry['generation']) && ($entry['generation'] > 0)) {
 169                                 $generation = ++$entry['generation'];
 170                         }
 171
 172                         if (isset($entry['tags'])) {
 173                                 foreach ($entry['tags'] as $tag) {
 174                                         $keywords = implode(", ", $tag);
 175                                 }
 176                         }
 177
 178                         if (isset($entry['contactType']) && ($entry['contactType'] >= 0)) {
 179                                 $contact_type = $entry['contactType'];
 180                         }
 181
 182                         $gcontact = ["url" => $profile_url,
 183                                         "name" => $name,
 184                                         "network" => $network,
 185                                         "photo" => $profile_photo,
 186                                         "about" => $about,
 187                                         "location" => $location,
 188                                         "gender" => $gender,
 189                                         "keywords" => $keywords,
 190                                         "connect" => $connect_url,
 191                                         "updated" => $updated,
 192                                         "contact-type" => $contact_type,
 193                                         "generation" => $generation];
 194
 195                         try {
 196                                 $gcontact = GContact::sanitize($gcontact);
 197                                 $gcid = GContact::update($gcontact);
 198
 199                                 GContact::link($gcid, $uid, $cid, $zcid);
 200                         } catch (Exception $e) {
 201                                 logger($e->getMessage(), LOGGER_DEBUG);
 202                         }
 203                 }
 204                 logger("load: loaded $total entries", LOGGER_DEBUG);
 205
 206                 $condition = ["`cid` = ? AND `uid` = ? AND `zcid` = ? AND `updated` < UTC_TIMESTAMP - INTERVAL 2 DAY", $cid, $uid, $zcid];
 207                 DBA::delete('glink', $condition);
 208         }
 209
 210         public static function reachable($profile, $server = "", $network = "", $force = false)
 211         {
 212                 if ($server == "") {
 213                         $server = self::detectServer($profile);
 214                 }
 215
 216                 if ($server == "") {
 217                         return true;
 218                 }
 219
 220                 return self::checkServer($server, $network, $force);
 221         }
 222
 223         public static function detectServer($profile)
 224         {
 225                 // Try to detect the server path based upon some known standard paths
 226                 $server_url = "";
 227
 228                 if ($server_url == "") {
 229                         $friendica = preg_replace("=(https?://)(.*)/profile/(.*)=ism", "$1$2", $profile);
 230                         if ($friendica != $profile) {
 231                                 $server_url = $friendica;
 232                                 $network = Protocol::DFRN;
 233                         }
 234                 }
 235
 236                 if ($server_url == "") {
 237                         $diaspora = preg_replace("=(https?://)(.*)/u/(.*)=ism", "$1$2", $profile);
 238                         if ($diaspora != $profile) {
 239                                 $server_url = $diaspora;
 240                                 $network = Protocol::DIASPORA;
 241                         }
 242                 }
 243
 244                 if ($server_url == "") {
 245                         $red = preg_replace("=(https?://)(.*)/channel/(.*)=ism", "$1$2", $profile);
 246                         if ($red != $profile) {
 247                                 $server_url = $red;
 248                                 $network = Protocol::DIASPORA;
 249                         }
 250                 }
 251
 252                 // Mastodon
 253                 if ($server_url == "") {
 254                         $mastodon = preg_replace("=(https?://)(.*)/users/(.*)=ism", "$1$2", $profile);
 255                         if ($mastodon != $profile) {
 256                                 $server_url = $mastodon;
 257                                 $network = Protocol::OSTATUS;
 258                         }
 259                 }
 260
 261                 // Numeric OStatus variant
 262                 if ($server_url == "") {
 263                         $ostatus = preg_replace("=(https?://)(.*)/user/(.*)=ism", "$1$2", $profile);
 264                         if ($ostatus != $profile) {
 265                                 $server_url = $ostatus;
 266                                 $network = Protocol::OSTATUS;
 267                         }
 268                 }
 269
 270                 // Wild guess
 271                 if ($server_url == "") {
 272                         $base = preg_replace("=(https?://)(.*?)/(.*)=ism", "$1$2", $profile);
 273                         if ($base != $profile) {
 274                                 $server_url = $base;
 275                                 $network = Protocol::PHANTOM;
 276                         }
 277                 }
 278
 279                 if ($server_url == "") {
 280                         return "";
 281                 }
 282
 283                 $r = q(
 284                         "SELECT `id` FROM `gserver` WHERE `nurl` = '%s' AND `last_contact` > `last_failure`",
 285                         DBA::escape(normalise_link($server_url))
 286                 );
 287
 288                 if (DBA::isResult($r)) {
 289                         return $server_url;
 290                 }
 291
 292                 // Fetch the host-meta to check if this really is a server
 293                 $serverret = Network::curl($server_url."/.well-known/host-meta");
 294                 if (!$serverret["success"]) {
 295                         return "";
 296                 }
 297
 298                 return $server_url;
 299         }
 300
 301         public static function alternateOStatusUrl($url)
 302         {
 303                 return(preg_match("=https?://.+/user/\d+=ism", $url, $matches));
 304         }
 305
 306         public static function lastUpdated($profile, $force = false)
 307         {
 308                 $gcontacts = q(
 309                         "SELECT * FROM `gcontact` WHERE `nurl` = '%s'",
 310                         DBA::escape(normalise_link($profile))
 311                 );
 312
 313                 if (!DBA::isResult($gcontacts)) {
 314                         return false;
 315                 }
 316
 317                 $contact = ["url" => $profile];
 318
 319                 if ($gcontacts[0]["created"] <= NULL_DATE) {
 320                         $contact['created'] = DateTimeFormat::utcNow();
 321                 }
 322
 323                 $server_url = '';
 324                 if ($force) {
 325                         $server_url = normalise_link(self::detectServer($profile));
 326                 }
 327
 328                 if (($server_url == '') && ($gcontacts[0]["server_url"] != "")) {
 329                         $server_url = $gcontacts[0]["server_url"];
 330                 }
 331
 332                 if (!$force && (($server_url == '') || ($gcontacts[0]["server_url"] == $gcontacts[0]["nurl"]))) {
 333                         $server_url = normalise_link(self::detectServer($profile));
 334                 }
 335
 336                 if (!in_array($gcontacts[0]["network"], [Protocol::DFRN, Protocol::DIASPORA, Protocol::FEED, Protocol::OSTATUS, ""])) {
 337                         logger("Profile ".$profile.": Network type ".$gcontacts[0]["network"]." can't be checked", LOGGER_DEBUG);
 338                         return false;
 339                 }
 340
 341                 if ($server_url != "") {
 342                         if (!self::checkServer($server_url, $gcontacts[0]["network"], $force)) {
 343                                 if ($force) {
 344                                         $fields = ['last_failure' => DateTimeFormat::utcNow()];
 345                                         DBA::update('gcontact', $fields, ['nurl' => normalise_link($profile)]);
 346                                 }
 347
 348                                 logger("Profile ".$profile.": Server ".$server_url." wasn't reachable.", LOGGER_DEBUG);
 349                                 return false;
 350                         }
 351                         $contact['server_url'] = $server_url;
 352                 }
 353
 354                 if (in_array($gcontacts[0]["network"], ["", Protocol::FEED])) {
 355                         $server = q(
 356                                 "SELECT `network` FROM `gserver` WHERE `nurl` = '%s' AND `network` != ''",
 357                                 DBA::escape(normalise_link($server_url))
 358                         );
 359
 360                         if ($server) {
 361                                 $contact['network'] = $server[0]["network"];
 362                         } else {
 363                                 return false;
 364                         }
 365                 }
 366
 367                 // noscrape is really fast so we don't cache the call.
 368                 if (($server_url != "") && ($gcontacts[0]["nick"] != "")) {
 369                         //  Use noscrape if possible
 370                         $server = q("SELECT `noscrape`, `network` FROM `gserver` WHERE `nurl` = '%s' AND `noscrape` != ''", DBA::escape(normalise_link($server_url)));
 371
 372                         if ($server) {
 373                                 $noscraperet = Network::curl($server[0]["noscrape"]."/".$gcontacts[0]["nick"]);
 374
 375                                 if ($noscraperet["success"] && ($noscraperet["body"] != "")) {
 376                                         $noscrape = json_decode($noscraperet["body"], true);
 377
 378                                         if (is_array($noscrape)) {
 379                                                 $contact["network"] = $server[0]["network"];
 380
 381                                                 if (isset($noscrape["fn"])) {
 382                                                         $contact["name"] = $noscrape["fn"];
 383                                                 }
 384                                                 if (isset($noscrape["comm"])) {
 385                                                         $contact["community"] = $noscrape["comm"];
 386                                                 }
 387                                                 if (isset($noscrape["tags"])) {
 388                                                         $keywords = implode(" ", $noscrape["tags"]);
 389                                                         if ($keywords != "") {
 390                                                                 $contact["keywords"] = $keywords;
 391                                                         }
 392                                                 }
 393
 394                                                 $location = Profile::formatLocation($noscrape);
 395                                                 if ($location) {
 396                                                         $contact["location"] = $location;
 397                                                 }
 398                                                 if (isset($noscrape["dfrn-notify"])) {
 399                                                         $contact["notify"] = $noscrape["dfrn-notify"];
 400                                                 }
 401                                                 // Remove all fields that are not present in the gcontact table
 402                                                 unset($noscrape["fn"]);
 403                                                 unset($noscrape["key"]);
 404                                                 unset($noscrape["homepage"]);
 405                                                 unset($noscrape["comm"]);
 406                                                 unset($noscrape["tags"]);
 407                                                 unset($noscrape["locality"]);
 408                                                 unset($noscrape["region"]);
 409                                                 unset($noscrape["country-name"]);
 410                                                 unset($noscrape["contacts"]);
 411                                                 unset($noscrape["dfrn-request"]);
 412                                                 unset($noscrape["dfrn-confirm"]);
 413                                                 unset($noscrape["dfrn-notify"]);
 414                                                 unset($noscrape["dfrn-poll"]);
 415
 416                                                 // Set the date of the last contact
 417                                                 /// @todo By now the function "update_gcontact" doesn't work with this field
 418                                                 //$contact["last_contact"] = DateTimeFormat::utcNow();
 419
 420                                                 $contact = array_merge($contact, $noscrape);
 421
 422                                                 GContact::update($contact);
 423
 424                                                 if (!empty($noscrape["updated"])) {
 425                                                         $fields = ['last_contact' => DateTimeFormat::utcNow()];
 426                                                         DBA::update('gcontact', $fields, ['nurl' => normalise_link($profile)]);
 427
 428                                                         logger("Profile ".$profile." was last updated at ".$noscrape["updated"]." (noscrape)", LOGGER_DEBUG);
 429
 430                                                         return $noscrape["updated"];
 431                                                 }
 432                                         }
 433                                 }
 434                         }
 435                 }
 436
 437                 // If we only can poll the feed, then we only do this once a while
 438                 if (!$force && !self::updateNeeded($gcontacts[0]["created"], $gcontacts[0]["updated"], $gcontacts[0]["last_failure"], $gcontacts[0]["last_contact"])) {
 439                         logger("Profile ".$profile." was last updated at ".$gcontacts[0]["updated"]." (cached)", LOGGER_DEBUG);
 440
 441                         GContact::update($contact);
 442                         return $gcontacts[0]["updated"];
 443                 }
 444
 445                 $data = Probe::uri($profile);
 446
 447                 // Is the profile link the alternate OStatus link notation? (http://domain.tld/user/4711)
 448                 // Then check the other link and delete this one
 449                 if (($data["network"] == Protocol::OSTATUS) && self::alternateOStatusUrl($profile)
 450                         && (normalise_link($profile) == normalise_link($data["alias"]))
 451                         && (normalise_link($profile) != normalise_link($data["url"]))
 452                 ) {
 453                         // Delete the old entry
 454                         DBA::delete('gcontact', ['nurl' => normalise_link($profile)]);
 455
 456                         $gcontact = array_merge($gcontacts[0], $data);
 457
 458                         $gcontact["server_url"] = $data["baseurl"];
 459
 460                         try {
 461                                 $gcontact = GContact::sanitize($gcontact);
 462                                 GContact::update($gcontact);
 463
 464                                 self::lastUpdated($data["url"], $force);
 465                         } catch (Exception $e) {
 466                                 logger($e->getMessage(), LOGGER_DEBUG);
 467                         }
 468
 469                         logger("Profile ".$profile." was deleted", LOGGER_DEBUG);
 470                         return false;
 471                 }
 472
 473                 if (($data["poll"] == "") || (in_array($data["network"], [Protocol::FEED, Protocol::PHANTOM]))) {
 474                         $fields = ['last_failure' => DateTimeFormat::utcNow()];
 475                         DBA::update('gcontact', $fields, ['nurl' => normalise_link($profile)]);
 476
 477                         logger("Profile ".$profile." wasn't reachable (profile)", LOGGER_DEBUG);
 478                         return false;
 479                 }
 480
 481                 $contact = array_merge($contact, $data);
 482
 483                 $contact["server_url"] = $data["baseurl"];
 484
 485                 GContact::update($contact);
 486
 487                 $feedret = Network::curl($data["poll"]);
 488
 489                 if (!$feedret["success"]) {
 490                         $fields = ['last_failure' => DateTimeFormat::utcNow()];
 491                         DBA::update('gcontact', $fields, ['nurl' => normalise_link($profile)]);
 492
 493                         logger("Profile ".$profile." wasn't reachable (no feed)", LOGGER_DEBUG);
 494                         return false;
 495                 }
 496
 497                 $doc = new DOMDocument();
 498                 /// @TODO Avoid error supression here
 499                 @$doc->loadXML($feedret["body"]);
 500
 501                 $xpath = new DOMXPath($doc);
 502                 $xpath->registerNamespace('atom', "http://www.w3.org/2005/Atom");
 503
 504                 $entries = $xpath->query('/atom:feed/atom:entry');
 505
 506                 $last_updated = "";
 507
 508                 foreach ($entries as $entry) {
 509                         $published = DateTimeFormat::utc($xpath->query('atom:published/text()', $entry)->item(0)->nodeValue);
 510                         $updated   = DateTimeFormat::utc($xpath->query('atom:updated/text()'  , $entry)->item(0)->nodeValue);
 511
 512                         if ($last_updated < $published) {
 513                                 $last_updated = $published;
 514                         }
 515
 516                         if ($last_updated < $updated) {
 517                                 $last_updated = $updated;
 518                         }
 519                 }
 520
 521                 // Maybe there aren't any entries. Then check if it is a valid feed
 522                 if ($last_updated == "") {
 523                         if ($xpath->query('/atom:feed')->length > 0) {
 524                                 $last_updated = NULL_DATE;
 525                         }
 526                 }
 527
 528                 $fields = ['last_contact' => DateTimeFormat::utcNow()];
 529
 530                 if (!empty($last_updated)) {
 531                         $fields['updated'] = $last_updated;
 532                 }
 533
 534                 DBA::update('gcontact', $fields, ['nurl' => normalise_link($profile)]);
 535
 536                 if (($gcontacts[0]["generation"] == 0)) {
 537                         $fields = ['generation' => 9];
 538                         DBA::update('gcontact', $fields, ['nurl' => normalise_link($profile)]);
 539                 }
 540
 541                 logger("Profile ".$profile." was last updated at ".$last_updated, LOGGER_DEBUG);
 542
 543                 return $last_updated;
 544         }
 545
 546         public static function updateNeeded($created, $updated, $last_failure, $last_contact)
 547         {
 548                 $now = strtotime(DateTimeFormat::utcNow());
 549
 550                 if ($updated > $last_contact) {
 551                         $contact_time = strtotime($updated);
 552                 } else {
 553                         $contact_time = strtotime($last_contact);
 554                 }
 555
 556                 $failure_time = strtotime($last_failure);
 557                 $created_time = strtotime($created);
 558
 559                 // If there is no "created" time then use the current time
 560                 if ($created_time <= 0) {
 561                         $created_time = $now;
 562                 }
 563
 564                 // If the last contact was less than 24 hours then don't update
 565                 if (($now - $contact_time) < (60 * 60 * 24)) {
 566                         return false;
 567                 }
 568
 569                 // If the last failure was less than 24 hours then don't update
 570                 if (($now - $failure_time) < (60 * 60 * 24)) {
 571                         return false;
 572                 }
 573
 574                 // If the last contact was less than a week ago and the last failure is older than a week then don't update
 575                 //if ((($now - $contact_time) < (60 * 60 * 24 * 7)) && ($contact_time > $failure_time))
 576                 //      return false;
 577
 578                 // If the last contact time was more than a week ago and the contact was created more than a week ago, then only try once a week
 579                 if ((($now - $contact_time) > (60 * 60 * 24 * 7)) && (($now - $created_time) > (60 * 60 * 24 * 7)) && (($now - $failure_time) < (60 * 60 * 24 * 7))) {
 580                         return false;
 581                 }
 582
 583                 // If the last contact time was more than a month ago and the contact was created more than a month ago, then only try once a month
 584                 if ((($now - $contact_time) > (60 * 60 * 24 * 30)) && (($now - $created_time) > (60 * 60 * 24 * 30)) && (($now - $failure_time) < (60 * 60 * 24 * 30))) {
 585                         return false;
 586                 }
 587
 588                 return true;
 589         }
 590
 591         /// @TODO Maybe move this out to an utilities class?
 592         private static function toBoolean($val)
 593         {
 594                 if (($val == "true") || ($val == 1)) {
 595                         return true;
 596                 } elseif (($val == "false") || ($val == 0)) {
 597                         return false;
 598                 }
 599
 600                 return $val;
 601         }
 602
 603         /**
 604          * @brief Detect server type (Hubzilla or Friendica) via the poco data
 605          *
 606          * @param array $data POCO data
 607          * @return array Server data
 608          */
 609         private static function detectPocoData(array $data)
 610         {
 611                 $server = false;
 612
 613                 if (!isset($data['entry'])) {
 614                         return false;
 615                 }
 616
 617                 if (count($data['entry']) == 0) {
 618                         return false;
 619                 }
 620
 621                 if (!isset($data['entry'][0]['urls'])) {
 622                         return false;
 623                 }
 624
 625                 if (count($data['entry'][0]['urls']) == 0) {
 626                         return false;
 627                 }
 628
 629                 foreach ($data['entry'][0]['urls'] as $url) {
 630                         if ($url['type'] == 'zot') {
 631                                 $server = [];
 632                                 $server["platform"] = 'Hubzilla';
 633                                 $server["network"] = Protocol::DIASPORA;
 634                                 return $server;
 635                         }
 636                 }
 637                 return false;
 638         }
 639
 640         /**
 641          * @brief Detect server type by using the nodeinfo data
 642          *
 643          * @param string $server_url address of the server
 644          * @return array Server data
 645          */
 646         private static function fetchNodeinfo($server_url)
 647         {
 648                 $serverret = Network::curl($server_url."/.well-known/nodeinfo");
 649                 if (!$serverret["success"]) {
 650                         return false;
 651                 }
 652
 653                 $nodeinfo = json_decode($serverret['body'], true);
 654
 655                 if (!is_array($nodeinfo) || !isset($nodeinfo['links'])) {
 656                         return false;
 657                 }
 658
 659                 $nodeinfo1_url = '';
 660                 $nodeinfo2_url = '';
 661
 662                 foreach ($nodeinfo['links'] as $link) {
 663                         if (!is_array($link) || empty($link['rel'])) {
 664                                 logger('Invalid nodeinfo format for ' . $server_url, LOGGER_DEBUG);
 665                                 continue;
 666                         }
 667                         if ($link['rel'] == 'http://nodeinfo.diaspora.software/ns/schema/1.0') {
 668                                 $nodeinfo1_url = $link['href'];
 669                         } elseif ($link['rel'] == 'http://nodeinfo.diaspora.software/ns/schema/2.0') {
 670                                 $nodeinfo2_url = $link['href'];
 671                         }
 672                 }
 673
 674                 if ($nodeinfo1_url . $nodeinfo2_url == '') {
 675                         return false;
 676                 }
 677
 678                 $server = [];
 679
 680                 // When the nodeinfo url isn't on the same host, then there is obviously something wrong
 681                 if (!empty($nodeinfo2_url) && (parse_url($server_url, PHP_URL_HOST) == parse_url($nodeinfo2_url, PHP_URL_HOST))) {
 682                         $server = self::parseNodeinfo2($nodeinfo2_url);
 683                 }
 684
 685                 // When the nodeinfo url isn't on the same host, then there is obviously something wrong
 686                 if (empty($server) && !empty($nodeinfo1_url) && (parse_url($server_url, PHP_URL_HOST) == parse_url($nodeinfo1_url, PHP_URL_HOST))) {
 687                         $server = self::parseNodeinfo1($nodeinfo1_url);
 688                 }
 689
 690                 return $server;
 691         }
 692
 693         /**
 694          * @brief Parses Nodeinfo 1
 695          *
 696          * @param string $nodeinfo_url address of the nodeinfo path
 697          * @return array Server data
 698          */
 699         private static function parseNodeinfo1($nodeinfo_url)
 700         {
 701                 $serverret = Network::curl($nodeinfo_url);
 702
 703                 if (!$serverret["success"]) {
 704                         return false;
 705                 }
 706
 707                 $nodeinfo = json_decode($serverret['body'], true);
 708
 709                 if (!is_array($nodeinfo)) {
 710                         return false;
 711                 }
 712
 713                 $server = [];
 714
 715                 $server['register_policy'] = REGISTER_CLOSED;
 716
 717                 if (is_bool($nodeinfo['openRegistrations']) && $nodeinfo['openRegistrations']) {
 718                         $server['register_policy'] = REGISTER_OPEN;
 719                 }
 720
 721                 if (is_array($nodeinfo['software'])) {
 722                         if (isset($nodeinfo['software']['name'])) {
 723                                 $server['platform'] = $nodeinfo['software']['name'];
 724                         }
 725
 726                         if (isset($nodeinfo['software']['version'])) {
 727                                 $server['version'] = $nodeinfo['software']['version'];
 728                                 // Version numbers on Nodeinfo are presented with additional info, e.g.:
 729                                 // 0.6.3.0-p1702cc1c, 0.6.99.0-p1b9ab160 or 3.4.3-2-1191.
 730                                 $server['version'] = preg_replace("=(.+)-(.{4,})=ism", "$1", $server['version']);
 731                         }
 732                 }
 733
 734                 if (is_array($nodeinfo['metadata']) && isset($nodeinfo['metadata']['nodeName'])) {
 735                         $server['site_name'] = $nodeinfo['metadata']['nodeName'];
 736                 }
 737
 738                 if (!empty($nodeinfo['usage']['users']['total'])) {
 739                         $server['registered-users'] = $nodeinfo['usage']['users']['total'];
 740                 }
 741
 742                 $diaspora = false;
 743                 $friendica = false;
 744                 $gnusocial = false;
 745
 746                 if (is_array($nodeinfo['protocols']['inbound'])) {
 747                         foreach ($nodeinfo['protocols']['inbound'] as $inbound) {
 748                                 if ($inbound == 'diaspora') {
 749                                         $diaspora = true;
 750                                 }
 751                                 if ($inbound == 'friendica') {
 752                                         $friendica = true;
 753                                 }
 754                                 if ($inbound == 'gnusocial') {
 755                                         $gnusocial = true;
 756                                 }
 757                         }
 758                 }
 759
 760                 if ($gnusocial) {
 761                         $server['network'] = Protocol::OSTATUS;
 762                 }
 763                 if ($diaspora) {
 764                         $server['network'] = Protocol::DIASPORA;
 765                 }
 766                 if ($friendica) {
 767                         $server['network'] = Protocol::DFRN;
 768                 }
 769
 770                 if (!$server) {
 771                         return false;
 772                 }
 773
 774                 return $server;
 775         }
 776
 777         /**
 778          * @brief Parses Nodeinfo 2
 779          *
 780          * @param string $nodeinfo_url address of the nodeinfo path
 781          * @return array Server data
 782          */
 783         private static function parseNodeinfo2($nodeinfo_url)
 784         {
 785                 $serverret = Network::curl($nodeinfo_url);
 786                 if (!$serverret["success"]) {
 787                         return false;
 788                 }
 789
 790                 $nodeinfo = json_decode($serverret['body'], true);
 791
 792                 if (!is_array($nodeinfo)) {
 793                         return false;
 794                 }
 795
 796                 $server = [];
 797
 798                 $server['register_policy'] = REGISTER_CLOSED;
 799
 800                 if (is_bool($nodeinfo['openRegistrations']) && $nodeinfo['openRegistrations']) {
 801                         $server['register_policy'] = REGISTER_OPEN;
 802                 }
 803
 804                 if (is_array($nodeinfo['software'])) {
 805                         if (isset($nodeinfo['software']['name'])) {
 806                                 $server['platform'] = $nodeinfo['software']['name'];
 807                         }
 808
 809                         if (isset($nodeinfo['software']['version'])) {
 810                                 $server['version'] = $nodeinfo['software']['version'];
 811                                 // Version numbers on Nodeinfo are presented with additional info, e.g.:
 812                                 // 0.6.3.0-p1702cc1c, 0.6.99.0-p1b9ab160 or 3.4.3-2-1191.
 813                                 $server['version'] = preg_replace("=(.+)-(.{4,})=ism", "$1", $server['version']);
 814                         }
 815                 }
 816
 817                 if (is_array($nodeinfo['metadata']) && isset($nodeinfo['metadata']['nodeName'])) {
 818                         $server['site_name'] = $nodeinfo['metadata']['nodeName'];
 819                 }
 820
 821                 if (!empty($nodeinfo['usage']['users']['total'])) {
 822                         $server['registered-users'] = $nodeinfo['usage']['users']['total'];
 823                 }
 824
 825                 $diaspora = false;
 826                 $friendica = false;
 827                 $gnusocial = false;
 828
 829                 if (!empty($nodeinfo['protocols'])) {
 830                         foreach ($nodeinfo['protocols'] as $protocol) {
 831                                 if ($protocol == 'diaspora') {
 832                                         $diaspora = true;
 833                                 } elseif ($protocol == 'friendica') {
 834                                         $friendica = true;
 835                                 } elseif ($protocol == 'gnusocial') {
 836                                         $gnusocial = true;
 837                                 }
 838                         }
 839                 }
 840
 841                 if ($gnusocial) {
 842                         $server['network'] = Protocol::OSTATUS;
 843                 } elseif ($diaspora) {
 844                         $server['network'] = Protocol::DIASPORA;
 845                 } elseif ($friendica) {
 846                         $server['network'] = Protocol::DFRN;
 847                 }
 848
 849                 if (empty($server)) {
 850                         return false;
 851                 }
 852
 853                 return $server;
 854         }
 855
 856         /**
 857          * @brief Detect server type (Hubzilla or Friendica) via the front page body
 858          *
 859          * @param string $body Front page of the server
 860          * @return array Server data
 861          */
 862         private static function detectServerType($body)
 863         {
 864                 $server = false;
 865
 866                 $doc = new DOMDocument();
 867                 /// @TODO Acoid supressing error
 868                 @$doc->loadHTML($body);
 869                 $xpath = new DOMXPath($doc);
 870
 871                 $list = $xpath->query("//meta[@name]");
 872
 873                 foreach ($list as $node) {
 874                         $attr = [];
 875                         if ($node->attributes->length) {
 876                                 foreach ($node->attributes as $attribute) {
 877                                         $attr[$attribute->name] = $attribute->value;
 878                                 }
 879                         }
 880                         if ($attr['name'] == 'generator') {
 881                                 $version_part = explode(" ", $attr['content']);
 882                                 if (count($version_part) == 2) {
 883                                         if (in_array($version_part[0], ["Friendika", "Friendica"])) {
 884                                                 $server = [];
 885                                                 $server["platform"] = $version_part[0];
 886                                                 $server["version"] = $version_part[1];
 887                                                 $server["network"] = Protocol::DFRN;
 888                                         }
 889                                 }
 890                         }
 891                 }
 892
 893                 if (!$server) {
 894                         $list = $xpath->query("//meta[@property]");
 895
 896                         foreach ($list as $node) {
 897                                 $attr = [];
 898                                 if ($node->attributes->length) {
 899                                         foreach ($node->attributes as $attribute) {
 900                                                 $attr[$attribute->name] = $attribute->value;
 901                                         }
 902                                 }
 903                                 if ($attr['property'] == 'generator' && in_array($attr['content'], ["hubzilla", "BlaBlaNet"])) {
 904                                         $server = [];
 905                                         $server["platform"] = $attr['content'];
 906                                         $server["version"] = "";
 907                                         $server["network"] = Protocol::DIASPORA;
 908                                 }
 909                         }
 910                 }
 911
 912                 if (!$server) {
 913                         return false;
 914                 }
 915
 916                 $server["site_name"] = XML::getFirstNodeValue($xpath, '//head/title/text()');
 917
 918                 return $server;
 919         }
 920
 921         public static function checkServer($server_url, $network = "", $force = false)
 922         {
 923                 // Unify the server address
 924                 $server_url = trim($server_url, "/");
 925                 $server_url = str_replace("/index.php", "", $server_url);
 926
 927                 if ($server_url == "") {
 928                         return false;
 929                 }
 930
 931                 $gserver = DBA::selectFirst('gserver', [], ['nurl' => normalise_link($server_url)]);
 932                 if (DBA::isResult($gserver)) {
 933                         if ($gserver["created"] <= NULL_DATE) {
 934                                 $fields = ['created' => DateTimeFormat::utcNow()];
 935                                 $condition = ['nurl' => normalise_link($server_url)];
 936                                 DBA::update('gserver', $fields, $condition);
 937                         }
 938                         $poco = $gserver["poco"];
 939                         $noscrape = $gserver["noscrape"];
 940
 941                         if ($network == "") {
 942                                 $network = $gserver["network"];
 943                         }
 944
 945                         $last_contact = $gserver["last_contact"];
 946                         $last_failure = $gserver["last_failure"];
 947                         $version = $gserver["version"];
 948                         $platform = $gserver["platform"];
 949                         $site_name = $gserver["site_name"];
 950                         $info = $gserver["info"];
 951                         $register_policy = $gserver["register_policy"];
 952                         $registered_users = $gserver["registered-users"];
 953
 954                         // See discussion under https://forum.friendi.ca/display/0b6b25a8135aabc37a5a0f5684081633
 955                         // It can happen that a zero date is in the database, but storing it again is forbidden.
 956                         if ($last_contact < NULL_DATE) {
 957                                 $last_contact = NULL_DATE;
 958                         }
 959
 960                         if ($last_failure < NULL_DATE) {
 961                                 $last_failure = NULL_DATE;
 962                         }
 963
 964                         if (!$force && !self::updateNeeded($gserver["created"], "", $last_failure, $last_contact)) {
 965                                 logger("Use cached data for server ".$server_url, LOGGER_DEBUG);
 966                                 return ($last_contact >= $last_failure);
 967                         }
 968                 } else {
 969                         $poco = "";
 970                         $noscrape = "";
 971                         $version = "";
 972                         $platform = "";
 973                         $site_name = "";
 974                         $info = "";
 975                         $register_policy = -1;
 976                         $registered_users = 0;
 977
 978                         $last_contact = NULL_DATE;
 979                         $last_failure = NULL_DATE;
 980                 }
 981                 logger("Server ".$server_url." is outdated or unknown. Start discovery. Force: ".$force." Created: ".$gserver["created"]." Failure: ".$last_failure." Contact: ".$last_contact, LOGGER_DEBUG);
 982
 983                 $failure = false;
 984                 $possible_failure = false;
 985                 $orig_last_failure = $last_failure;
 986                 $orig_last_contact = $last_contact;
 987
 988                 // Mastodon uses the "@" for user profiles.
 989                 // But this can be misunderstood.
 990                 if (parse_url($server_url, PHP_URL_USER) != '') {
 991                         DBA::update('gserver', ['last_failure' => DateTimeFormat::utcNow()], ['nurl' => normalise_link($server_url)]);
 992                         return false;
 993                 }
 994
 995                 // Check if the page is accessible via SSL.
 996                 $orig_server_url = $server_url;
 997                 $server_url = str_replace("http://", "https://", $server_url);
 998
 999                 // We set the timeout to 20 seconds since this operation should be done in no time if the server was vital
1000                 $serverret = Network::curl($server_url."/.well-known/host-meta", false, $redirects, ['timeout' => 20]);
1001
1002                 // Quit if there is a timeout.
1003                 // But we want to make sure to only quit if we are mostly sure that this server url fits.
1004                 if (DBA::isResult($gserver) && ($orig_server_url == $server_url) &&
1005                         (!empty($serverret["errno"]) && ($serverret['errno'] == CURLE_OPERATION_TIMEDOUT))) {
1006                         logger("Connection to server ".$server_url." timed out.", LOGGER_DEBUG);
1007                         DBA::update('gserver', ['last_failure' => DateTimeFormat::utcNow()], ['nurl' => normalise_link($server_url)]);
1008                         return false;
1009                 }
1010
1011                 // Maybe the page is unencrypted only?
1012                 $xmlobj = @simplexml_load_string($serverret["body"], 'SimpleXMLElement', 0, "http://docs.oasis-open.org/ns/xri/xrd-1.0");
1013                 if (!$serverret["success"] || ($serverret["body"] == "") || empty($xmlobj) || !is_object($xmlobj)) {
1014                         $server_url = str_replace("https://", "http://", $server_url);
1015
1016                         // We set the timeout to 20 seconds since this operation should be done in no time if the server was vital
1017                         $serverret = Network::curl($server_url."/.well-known/host-meta", false, $redirects, ['timeout' => 20]);
1018
1019                         // Quit if there is a timeout
1020                         if (!empty($serverret["errno"]) && ($serverret['errno'] == CURLE_OPERATION_TIMEDOUT)) {
1021                                 logger("Connection to server ".$server_url." timed out.", LOGGER_DEBUG);
1022                                 DBA::update('gserver', ['last_failure' => DateTimeFormat::utcNow()], ['nurl' => normalise_link($server_url)]);
1023                                 return false;
1024                         }
1025
1026                         $xmlobj = @simplexml_load_string($serverret["body"], 'SimpleXMLElement', 0, "http://docs.oasis-open.org/ns/xri/xrd-1.0");
1027                 }
1028
1029                 if (!$serverret["success"] || ($serverret["body"] == "") || empty($xmlobj) || !is_object($xmlobj)) {
1030                         // Workaround for bad configured servers (known nginx problem)
1031                         if (!empty($serverret["debug"]) && !in_array($serverret["debug"]["http_code"], ["403", "404"])) {
1032                                 $failure = true;
1033                         }
1034
1035                         $possible_failure = true;
1036                 }
1037
1038                 // If the server has no possible failure we reset the cached data
1039                 if (!$possible_failure) {
1040                         $version = "";
1041                         $platform = "";
1042                         $site_name = "";
1043                         $info = "";
1044                         $register_policy = -1;
1045                 }
1046
1047                 if (!$failure) {
1048                         // This will be too low, but better than no value at all.
1049                         $registered_users = DBA::count('gcontact', ['server_url' => normalise_link($server_url)]);
1050                 }
1051
1052                 // Look for poco
1053                 if (!$failure) {
1054                         $serverret = Network::curl($server_url."/poco");
1055
1056                         if ($serverret["success"]) {
1057                                 $data = json_decode($serverret["body"], true);
1058
1059                                 if (isset($data['totalResults'])) {
1060                                         $registered_users = $data['totalResults'];
1061                                         $poco = $server_url . "/poco";
1062                                         $server = self::detectPocoData($data);
1063
1064                                         if (!empty($server)) {
1065                                                 $platform = $server['platform'];
1066                                                 $network = $server['network'];
1067                                                 $version = '';
1068                                                 $site_name = '';
1069                                         }
1070                                 }
1071
1072                                 /*
1073                                  * There are servers out there who don't return 404 on a failure
1074                                  * We have to be sure that don't misunderstand this
1075                                  */
1076                                 if (is_null($data)) {
1077                                         $poco = "";
1078                                         $noscrape = "";
1079                                         $network = "";
1080                                 }
1081                         }
1082                 }
1083
1084                 if (!$failure) {
1085                         // Test for Diaspora, Hubzilla, Mastodon or older Friendica servers
1086                         $serverret = Network::curl($server_url);
1087
1088                         if (!$serverret["success"] || ($serverret["body"] == "")) {
1089                                 $failure = true;
1090                         } else {
1091                                 $server = self::detectServerType($serverret["body"]);
1092
1093                                 if (!empty($server)) {
1094                                         $platform = $server['platform'];
1095                                         $network = $server['network'];
1096                                         $version = $server['version'];
1097                                         $site_name = $server['site_name'];
1098                                 }
1099
1100                                 $lines = explode("\n", $serverret["header"]);
1101
1102                                 if (count($lines)) {
1103                                         foreach ($lines as $line) {
1104                                                 $line = trim($line);
1105
1106                                                 if (stristr($line, 'X-Diaspora-Version:')) {
1107                                                         $platform = "Diaspora";
1108                                                         $version = trim(str_replace("X-Diaspora-Version:", "", $line));
1109                                                         $version = trim(str_replace("x-diaspora-version:", "", $version));
1110                                                         $network = Protocol::DIASPORA;
1111                                                         $versionparts = explode("-", $version);
1112                                                         $version = $versionparts[0];
1113                                                 }
1114
1115                                                 if (stristr($line, 'Server: Mastodon')) {
1116                                                         $platform = "Mastodon";
1117                                                         $network = Protocol::OSTATUS;
1118                                                 }
1119                                         }
1120                                 }
1121                         }
1122                 }
1123
1124                 if (!$failure && ($poco == "")) {
1125                         // Test for Statusnet
1126                         // Will also return data for Friendica and GNU Social - but it will be overwritten later
1127                         // The "not implemented" is a special treatment for really, really old Friendica versions
1128                         $serverret = Network::curl($server_url."/api/statusnet/version.json");
1129
1130                         if ($serverret["success"] && ($serverret["body"] != '{"error":"not implemented"}') &&
1131                                 ($serverret["body"] != '') && (strlen($serverret["body"]) < 30)) {
1132                                 $platform = "StatusNet";
1133                                 // Remove junk that some GNU Social servers return
1134                                 $version = str_replace(chr(239).chr(187).chr(191), "", $serverret["body"]);
1135                                 $version = trim($version, '"');
1136                                 $network = Protocol::OSTATUS;
1137                         }
1138
1139                         // Test for GNU Social
1140                         $serverret = Network::curl($server_url."/api/gnusocial/version.json");
1141
1142                         if ($serverret["success"] && ($serverret["body"] != '{"error":"not implemented"}') &&
1143                                 ($serverret["body"] != '') && (strlen($serverret["body"]) < 30)) {
1144                                 $platform = "GNU Social";
1145                                 // Remove junk that some GNU Social servers return
1146                                 $version = str_replace(chr(239) . chr(187) . chr(191), "", $serverret["body"]);
1147                                 $version = trim($version, '"');
1148                                 $network = Protocol::OSTATUS;
1149                         }
1150
1151                         // Test for Mastodon
1152                         $orig_version = $version;
1153                         $serverret = Network::curl($server_url . "/api/v1/instance");
1154
1155                         if ($serverret["success"] && ($serverret["body"] != '')) {
1156                                 $data = json_decode($serverret["body"], true);
1157
1158                                 if (isset($data['version'])) {
1159                                         $platform = "Mastodon";
1160                                         $version = defaults($data, 'version', '');
1161                                         $site_name = defaults($data, 'title', '');
1162                                         $info = defaults($data, 'description', '');
1163                                         $network = Protocol::OSTATUS;
1164                                 }
1165
1166                                 if (!empty($data['stats']['user_count'])) {
1167                                         $registered_users = $data['stats']['user_count'];
1168                                 }
1169                         }
1170
1171                         if (strstr($orig_version . $version, 'Pleroma')) {
1172                                 $platform = 'Pleroma';
1173                                 $version = trim(str_replace('Pleroma', '', $version));
1174                         }
1175                 }
1176
1177                 if (!$failure) {
1178                         // Test for Hubzilla and Red
1179                         $serverret = Network::curl($server_url . "/siteinfo.json");
1180
1181                         if ($serverret["success"]) {
1182                                 $data = json_decode($serverret["body"], true);
1183
1184                                 if (isset($data['url'])) {
1185                                         $platform = $data['platform'];
1186                                         $version = $data['version'];
1187                                         $network = Protocol::DIASPORA;
1188                                 }
1189
1190                                 if (!empty($data['site_name'])) {
1191                                         $site_name = $data['site_name'];
1192                                 }
1193
1194                                 if (!empty($data['channels_total'])) {
1195                                         $registered_users = $data['channels_total'];
1196                                 }
1197
1198                                 if (!empty($data['register_policy'])) {
1199                                         switch ($data['register_policy']) {
1200                                                 case "REGISTER_OPEN":
1201                                                         $register_policy = REGISTER_OPEN;
1202                                                         break;
1203
1204                                                 case "REGISTER_APPROVE":
1205                                                         $register_policy = REGISTER_APPROVE;
1206                                                         break;
1207
1208                                                 case "REGISTER_CLOSED":
1209                                                 default:
1210                                                         $register_policy = REGISTER_CLOSED;
1211                                                         break;
1212                                         }
1213                                 }
1214                         } else {
1215                                 // Test for Hubzilla, Redmatrix or Friendica
1216                                 $serverret = Network::curl($server_url."/api/statusnet/config.json");
1217
1218                                 if ($serverret["success"]) {
1219                                         $data = json_decode($serverret["body"], true);
1220
1221                                         if (isset($data['site']['server'])) {
1222                                                 if (isset($data['site']['platform'])) {
1223                                                         $platform = $data['site']['platform']['PLATFORM_NAME'];
1224                                                         $version = $data['site']['platform']['STD_VERSION'];
1225                                                         $network = Protocol::DIASPORA;
1226                                                 }
1227
1228                                                 if (isset($data['site']['BlaBlaNet'])) {
1229                                                         $platform = $data['site']['BlaBlaNet']['PLATFORM_NAME'];
1230                                                         $version = $data['site']['BlaBlaNet']['STD_VERSION'];
1231                                                         $network = Protocol::DIASPORA;
1232                                                 }
1233
1234                                                 if (isset($data['site']['hubzilla'])) {
1235                                                         $platform = $data['site']['hubzilla']['PLATFORM_NAME'];
1236                                                         $version = $data['site']['hubzilla']['RED_VERSION'];
1237                                                         $network = Protocol::DIASPORA;
1238                                                 }
1239
1240                                                 if (isset($data['site']['redmatrix'])) {
1241                                                         if (isset($data['site']['redmatrix']['PLATFORM_NAME'])) {
1242                                                                 $platform = $data['site']['redmatrix']['PLATFORM_NAME'];
1243                                                         } elseif (isset($data['site']['redmatrix']['RED_PLATFORM'])) {
1244                                                                 $platform = $data['site']['redmatrix']['RED_PLATFORM'];
1245                                                         }
1246
1247                                                         $version = $data['site']['redmatrix']['RED_VERSION'];
1248                                                         $network = Protocol::DIASPORA;
1249                                                 }
1250
1251                                                 if (isset($data['site']['friendica'])) {
1252                                                         $platform = $data['site']['friendica']['FRIENDICA_PLATFORM'];
1253                                                         $version = $data['site']['friendica']['FRIENDICA_VERSION'];
1254                                                         $network = Protocol::DFRN;
1255                                                 }
1256
1257                                                 $site_name = $data['site']['name'];
1258
1259                                                 $private = false;
1260                                                 $inviteonly = false;
1261                                                 $closed = false;
1262
1263                                                 if (!empty($data['site']['closed'])) {
1264                                                         $closed = self::toBoolean($data['site']['closed']);
1265                                                 }
1266
1267                                                 if (!empty($data['site']['private'])) {
1268                                                         $private = self::toBoolean($data['site']['private']);
1269                                                 }
1270
1271                                                 if (!empty($data['site']['inviteonly'])) {
1272                                                         $inviteonly = self::toBoolean($data['site']['inviteonly']);
1273                                                 }
1274
1275                                                 if (!$closed && !$private and $inviteonly) {
1276                                                         $register_policy = REGISTER_APPROVE;
1277                                                 } elseif (!$closed && !$private) {
1278                                                         $register_policy = REGISTER_OPEN;
1279                                                 } else {
1280                                                         $register_policy = REGISTER_CLOSED;
1281                                                 }
1282                                         }
1283                                 }
1284                         }
1285                 }
1286
1287                 // Query statistics.json. Optional package for Diaspora, Friendica and Redmatrix
1288                 if (!$failure) {
1289                         $serverret = Network::curl($server_url . "/statistics.json");
1290
1291                         if ($serverret["success"]) {
1292                                 $data = json_decode($serverret["body"], true);
1293
1294                                 if (isset($data['version'])) {
1295                                         $version = $data['version'];
1296                                         // Version numbers on statistics.json are presented with additional info, e.g.:
1297                                         // 0.6.3.0-p1702cc1c, 0.6.99.0-p1b9ab160 or 3.4.3-2-1191.
1298                                         $version = preg_replace("=(.+)-(.{4,})=ism", "$1", $version);
1299                                 }
1300
1301                                 if (!empty($data['name'])) {
1302                                         $site_name = $data['name'];
1303                                 }
1304
1305                                 if (!empty($data['network'])) {
1306                                         $platform = $data['network'];
1307                                 }
1308
1309                                 if ($platform == "Diaspora") {
1310                                         $network = Protocol::DIASPORA;
1311                                 }
1312
1313                                 if (!empty($data['registrations_open']) && $data['registrations_open']) {
1314                                         $register_policy = REGISTER_OPEN;
1315                                 } else {
1316                                         $register_policy = REGISTER_CLOSED;
1317                                 }
1318                         }
1319                 }
1320
1321                 // Query nodeinfo. Working for (at least) Diaspora and Friendica.
1322                 if (!$failure) {
1323                         $server = self::fetchNodeinfo($server_url);
1324
1325                         if (!empty($server)) {
1326                                 $register_policy = $server['register_policy'];
1327
1328                                 if (isset($server['platform'])) {
1329                                         $platform = $server['platform'];
1330                                 }
1331
1332                                 if (isset($server['network'])) {
1333                                         $network = $server['network'];
1334                                 }
1335
1336                                 if (isset($server['version'])) {
1337                                         $version = $server['version'];
1338                                 }
1339
1340                                 if (isset($server['site_name'])) {
1341                                         $site_name = $server['site_name'];
1342                                 }
1343
1344                                 if (isset($server['registered-users'])) {
1345                                         $registered_users = $server['registered-users'];
1346                                 }
1347                         }
1348                 }
1349
1350                 // Check for noscrape
1351                 // Friendica servers could be detected as OStatus servers
1352                 if (!$failure && in_array($network, [Protocol::DFRN, Protocol::OSTATUS])) {
1353                         $serverret = Network::curl($server_url . "/friendica/json");
1354
1355                         if (!$serverret["success"]) {
1356                                 $serverret = Network::curl($server_url . "/friendika/json");
1357                         }
1358
1359                         if ($serverret["success"]) {
1360                                 $data = json_decode($serverret["body"], true);
1361
1362                                 if (isset($data['version'])) {
1363                                         $network = Protocol::DFRN;
1364
1365                                         if (!empty($data['no_scrape_url'])) {
1366                                                 $noscrape = $data['no_scrape_url'];
1367                                         }
1368
1369                                         $version = $data['version'];
1370
1371                                         if (!empty($data['site_name'])) {
1372                                                 $site_name = $data['site_name'];
1373                                         }
1374
1375                                         $info = $data['info'];
1376                                         $register_policy = constant($data['register_policy']);
1377                                         $platform = $data['platform'];
1378                                 }
1379                         }
1380                 }
1381
1382                 // Every server has got at least an admin account
1383                 if (!$failure && ($registered_users == 0)) {
1384                         $registered_users = 1;
1385                 }
1386
1387                 if ($possible_failure && !$failure) {
1388                         $failure = true;
1389                 }
1390
1391                 if ($failure) {
1392                         $last_contact = $orig_last_contact;
1393                         $last_failure = DateTimeFormat::utcNow();
1394                 } else {
1395                         $last_contact = DateTimeFormat::utcNow();
1396                         $last_failure = $orig_last_failure;
1397                 }
1398
1399                 if (($last_contact <= $last_failure) && !$failure) {
1400                         logger("Server ".$server_url." seems to be alive, but last contact wasn't set - could be a bug", LOGGER_DEBUG);
1401                 } elseif (($last_contact >= $last_failure) && $failure) {
1402                         logger("Server ".$server_url." seems to be dead, but last failure wasn't set - could be a bug", LOGGER_DEBUG);
1403                 }
1404
1405                 // Check again if the server exists
1406                 $found = DBA::exists('gserver', ['nurl' => normalise_link($server_url)]);
1407
1408                 $version = strip_tags($version);
1409                 $site_name = strip_tags($site_name);
1410                 $info = strip_tags($info);
1411                 $platform = strip_tags($platform);
1412
1413                 $fields = ['url' => $server_url, 'version' => $version,
1414                                 'site_name' => $site_name, 'info' => $info, 'register_policy' => $register_policy,
1415                                 'poco' => $poco, 'noscrape' => $noscrape, 'network' => $network,
1416                                 'platform' => $platform, 'registered-users' => $registered_users,
1417                                 'last_contact' => $last_contact, 'last_failure' => $last_failure];
1418
1419                 if ($found) {
1420                         DBA::update('gserver', $fields, ['nurl' => normalise_link($server_url)]);
1421                 } elseif (!$failure) {
1422                         $fields['nurl'] = normalise_link($server_url);
1423                         $fields['created'] = DateTimeFormat::utcNow();
1424                         DBA::insert('gserver', $fields);
1425                 }
1426
1427                 if (!$failure && in_array($fields['network'], [Protocol::DFRN, Protocol::DIASPORA])) {
1428                         self::discoverRelay($server_url);
1429                 }
1430
1431                 logger("End discovery for server " . $server_url, LOGGER_DEBUG);
1432
1433                 return !$failure;
1434         }
1435
1436         /**
1437          * @brief Fetch relay data from a given server url
1438          *
1439          * @param string $server_url address of the server
1440          */
1441         private static function discoverRelay($server_url)
1442         {
1443                 logger("Discover relay data for server " . $server_url, LOGGER_DEBUG);
1444
1445                 $serverret = Network::curl($server_url . "/.well-known/x-social-relay");
1446
1447                 if (!$serverret["success"]) {
1448                         return;
1449                 }
1450
1451                 $data = json_decode($serverret['body'], true);
1452
1453                 if (!is_array($data)) {
1454                         return;
1455                 }
1456
1457                 $gserver = DBA::selectFirst('gserver', ['id', 'relay-subscribe', 'relay-scope'], ['nurl' => normalise_link($server_url)]);
1458
1459                 if (!DBA::isResult($gserver)) {
1460                         return;
1461                 }
1462
1463                 if (($gserver['relay-subscribe'] != $data['subscribe']) || ($gserver['relay-scope'] != $data['scope'])) {
1464                         $fields = ['relay-subscribe' => $data['subscribe'], 'relay-scope' => $data['scope']];
1465                         DBA::update('gserver', $fields, ['id' => $gserver['id']]);
1466                 }
1467
1468                 DBA::delete('gserver-tag', ['gserver-id' => $gserver['id']]);
1469
1470                 if ($data['scope'] == 'tags') {
1471                         // Avoid duplicates
1472                         $tags = [];
1473                         foreach ($data['tags'] as $tag) {
1474                                 $tag = mb_strtolower($tag);
1475                                 if (strlen($tag) < 100) {
1476                                         $tags[$tag] = $tag;
1477                                 }
1478                         }
1479
1480                         foreach ($tags as $tag) {
1481                                 DBA::insert('gserver-tag', ['gserver-id' => $gserver['id'], 'tag' => $tag], true);
1482                         }
1483                 }
1484
1485                 // Create or update the relay contact
1486                 $fields = [];
1487                 if (isset($data['protocols'])) {
1488                         if (isset($data['protocols']['diaspora'])) {
1489                                 $fields['network'] = Protocol::DIASPORA;
1490
1491                                 if (isset($data['protocols']['diaspora']['receive'])) {
1492                                         $fields['batch'] = $data['protocols']['diaspora']['receive'];
1493                                 } elseif (is_string($data['protocols']['diaspora'])) {
1494                                         $fields['batch'] = $data['protocols']['diaspora'];
1495                                 }
1496                         }
1497
1498                         if (isset($data['protocols']['dfrn'])) {
1499                                 $fields['network'] = Protocol::DFRN;
1500
1501                                 if (isset($data['protocols']['dfrn']['receive'])) {
1502                                         $fields['batch'] = $data['protocols']['dfrn']['receive'];
1503                                 } elseif (is_string($data['protocols']['dfrn'])) {
1504                                         $fields['batch'] = $data['protocols']['dfrn'];
1505                                 }
1506                         }
1507                 }
1508                 Diaspora::setRelayContact($server_url, $fields);
1509         }
1510
1511         /**
1512          * @brief Returns a list of all known servers
1513          * @return array List of server urls
1514          */
1515         public static function serverlist()
1516         {
1517                 $r = q(
1518                         "SELECT `url`, `site_name` AS `displayName`, `network`, `platform`, `version` FROM `gserver`
1519                         WHERE `network` IN ('%s', '%s', '%s') AND `last_contact` > `last_failure`
1520                         ORDER BY `last_contact`
1521                         LIMIT 1000",
1522                         DBA::escape(Protocol::DFRN),
1523                         DBA::escape(Protocol::DIASPORA),
1524                         DBA::escape(Protocol::OSTATUS)
1525                 );
1526
1527                 if (!DBA::isResult($r)) {
1528                         return false;
1529                 }
1530
1531                 return $r;
1532         }
1533
1534         /**
1535          * @brief Fetch server list from remote servers and adds them when they are new.
1536          *
1537          * @param string $poco URL to the POCO endpoint
1538          */
1539         private static function fetchServerlist($poco)
1540         {
1541                 $serverret = Network::curl($poco . "/@server");
1542
1543                 if (!$serverret["success"]) {
1544                         return;
1545                 }
1546
1547                 $serverlist = json_decode($serverret['body'], true);
1548
1549                 if (!is_array($serverlist)) {
1550                         return;
1551                 }
1552
1553                 foreach ($serverlist as $server) {
1554                         $server_url = str_replace("/index.php", "", $server['url']);
1555
1556                         $r = q("SELECT `nurl` FROM `gserver` WHERE `nurl` = '%s'", DBA::escape(normalise_link($server_url)));
1557
1558                         if (!DBA::isResult($r)) {
1559                                 logger("Call server check for server ".$server_url, LOGGER_DEBUG);
1560                                 Worker::add(PRIORITY_LOW, "DiscoverPoCo", "server", $server_url);
1561                         }
1562                 }
1563         }
1564
1565         private static function discoverFederation()
1566         {
1567                 $last = Config::get('poco', 'last_federation_discovery');
1568
1569                 if ($last) {
1570                         $next = $last + (24 * 60 * 60);
1571
1572                         if ($next > time()) {
1573                                 return;
1574                         }
1575                 }
1576
1577                 // Discover Friendica, Hubzilla and Diaspora servers
1578                 $serverdata = Network::fetchUrl("http://the-federation.info/pods.json");
1579
1580                 if (!empty($serverdata)) {
1581                         $servers = json_decode($serverdata, true);
1582
1583                         if (!empty($servers['pods'])) {
1584                                 foreach ($servers['pods'] as $server) {
1585                                         Worker::add(PRIORITY_LOW, "DiscoverPoCo", "server", "https://" . $server['host']);
1586                                 }
1587                         }
1588                 }
1589
1590                 // Disvover Mastodon servers
1591                 if (!Config::get('system', 'ostatus_disabled')) {
1592                         $accesstoken = Config::get('system', 'instances_social_key');
1593
1594                         if (!empty($accesstoken)) {
1595                                 $api = 'https://instances.social/api/1.0/instances/list?count=0';
1596                                 $header = ['Authorization: Bearer '.$accesstoken];
1597                                 $serverdata = Network::curl($api, false, $redirects, ['headers' => $header]);
1598
1599                                 if ($serverdata['success']) {
1600                                         $servers = json_decode($serverdata['body'], true);
1601
1602                                         foreach ($servers['instances'] as $server) {
1603                                                 $url = (is_null($server['https_score']) ? 'http' : 'https') . '://' . $server['name'];
1604                                                 Worker::add(PRIORITY_LOW, "DiscoverPoCo", "server", $url);
1605                                         }
1606                                 }
1607                         }
1608                 }
1609
1610                 // Currently disabled, since the service isn't available anymore.
1611                 // It is not removed since I hope that there will be a successor.
1612                 // Discover GNU Social Servers.
1613                 //if (!Config::get('system','ostatus_disabled')) {
1614                 //      $serverdata = "http://gstools.org/api/get_open_instances/";
1615
1616                 //      $result = Network::curl($serverdata);
1617                 //      if ($result["success"]) {
1618                 //              $servers = json_decode($result["body"], true);
1619
1620                 //              foreach($servers['data'] as $server)
1621                 //                      self::checkServer($server['instance_address']);
1622                 //      }
1623                 //}
1624
1625                 Config::set('poco', 'last_federation_discovery', time());
1626         }
1627
1628         public static function discoverSingleServer($id)
1629         {
1630                 $r = q("SELECT `poco`, `nurl`, `url`, `network` FROM `gserver` WHERE `id` = %d", intval($id));
1631
1632                 if (!DBA::isResult($r)) {
1633                         return false;
1634                 }
1635
1636                 $server = $r[0];
1637
1638                 // Discover new servers out there (Works from Friendica version 3.5.2)
1639                 self::fetchServerlist($server["poco"]);
1640
1641                 // Fetch all users from the other server
1642                 $url = $server["poco"] . "/?fields=displayName,urls,photos,updated,network,aboutMe,currentLocation,tags,gender,contactType,generation";
1643
1644                 logger("Fetch all users from the server " . $server["url"], LOGGER_DEBUG);
1645
1646                 $retdata = Network::curl($url);
1647
1648                 if ($retdata["success"] && !empty($retdata["body"])) {
1649                         $data = json_decode($retdata["body"], true);
1650
1651                         if (!empty($data)) {
1652                                 self::discoverServer($data, 2);
1653                         }
1654
1655                         if (Config::get('system', 'poco_discovery') > 1) {
1656                                 $timeframe = Config::get('system', 'poco_discovery_since');
1657
1658                                 if ($timeframe == 0) {
1659                                         $timeframe = 30;
1660                                 }
1661
1662                                 $updatedSince = date(DateTimeFormat::MYSQL, time() - $timeframe * 86400);
1663
1664                                 // Fetch all global contacts from the other server (Not working with Redmatrix and Friendica versions before 3.3)
1665                                 $url = $server["poco"]."/@global?updatedSince=".$updatedSince."&fields=displayName,urls,photos,updated,network,aboutMe,currentLocation,tags,gender,contactType,generation";
1666
1667                                 $success = false;
1668
1669                                 $retdata = Network::curl($url);
1670
1671                                 if ($retdata["success"] && !empty($retdata["body"])) {
1672                                         logger("Fetch all global contacts from the server " . $server["nurl"], LOGGER_DEBUG);
1673                                         $data = json_decode($retdata["body"], true);
1674
1675                                         if (!empty($data)) {
1676                                                 $success = self::discoverServer($data);
1677                                         }
1678                                 }
1679
1680                                 if (!$success && (Config::get('system', 'poco_discovery') > 2)) {
1681                                         logger("Fetch contacts from users of the server " . $server["nurl"], LOGGER_DEBUG);
1682                                         self::discoverServerUsers($data, $server);
1683                                 }
1684                         }
1685
1686                         $fields = ['last_poco_query' => DateTimeFormat::utcNow()];
1687                         DBA::update('gserver', $fields, ['nurl' => $server["nurl"]]);
1688
1689                         return true;
1690                 } else {
1691                         // If the server hadn't replied correctly, then force a sanity check
1692                         self::checkServer($server["url"], $server["network"], true);
1693
1694                         // If we couldn't reach the server, we will try it some time later
1695                         $fields = ['last_poco_query' => DateTimeFormat::utcNow()];
1696                         DBA::update('gserver', $fields, ['nurl' => $server["nurl"]]);
1697
1698                         return false;
1699                 }
1700         }
1701
1702         public static function discover($complete = false)
1703         {
1704                 // Update the server list
1705                 self::discoverFederation();
1706
1707                 $no_of_queries = 5;
1708
1709                 $requery_days = intval(Config::get('system', 'poco_requery_days'));
1710
1711                 if ($requery_days == 0) {
1712                         $requery_days = 7;
1713                 }
1714
1715                 $last_update = date('c', time() - (60 * 60 * 24 * $requery_days));
1716
1717                 $gservers = q("SELECT `id`, `url`, `nurl`, `network`
1718                         FROM `gserver`
1719                         WHERE `last_contact` >= `last_failure`
1720                         AND `poco` != ''
1721                         AND `last_poco_query` < '%s'
1722                         ORDER BY RAND()", DBA::escape($last_update)
1723                 );
1724
1725                 if (DBA::isResult($gservers)) {
1726                         foreach ($gservers as $gserver) {
1727                                 if (!self::checkServer($gserver['url'], $gserver['network'])) {
1728                                         // The server is not reachable? Okay, then we will try it later
1729                                         $fields = ['last_poco_query' => DateTimeFormat::utcNow()];
1730                                         DBA::update('gserver', $fields, ['nurl' => $gserver['nurl']]);
1731                                         continue;
1732                                 }
1733
1734                                 logger('Update directory from server ' . $gserver['url'] . ' with ID ' . $gserver['id'], LOGGER_DEBUG);
1735                                 Worker::add(PRIORITY_LOW, 'DiscoverPoCo', 'update_server_directory', (int) $gserver['id']);
1736
1737                                 if (!$complete && ( --$no_of_queries == 0)) {
1738                                         break;
1739                                 }
1740                         }
1741                 }
1742         }
1743
1744         private static function discoverServerUsers(array $data, array $server)
1745         {
1746                 if (!isset($data['entry'])) {
1747                         return;
1748                 }
1749
1750                 foreach ($data['entry'] as $entry) {
1751                         $username = '';
1752
1753                         if (isset($entry['urls'])) {
1754                                 foreach ($entry['urls'] as $url) {
1755                                         if ($url['type'] == 'profile') {
1756                                                 $profile_url = $url['value'];
1757                                                 $path_array = explode('/', parse_url($profile_url, PHP_URL_PATH));
1758                                                 $username = end($path_array);
1759                                         }
1760                                 }
1761                         }
1762
1763                         if ($username != '') {
1764                                 logger('Fetch contacts for the user ' . $username . ' from the server ' . $server['nurl'], LOGGER_DEBUG);
1765
1766                                 // Fetch all contacts from a given user from the other server
1767                                 $url = $server['poco'] . '/' . $username . '/?fields=displayName,urls,photos,updated,network,aboutMe,currentLocation,tags,gender,contactType,generation';
1768
1769                                 $retdata = Network::curl($url);
1770
1771                                 if (!empty($retdata['success'])) {
1772                                         $data = json_decode($retdata["body"], true);
1773
1774                                         if (!empty($data)) {
1775                                                 self::discoverServer($data, 3);
1776                                         }
1777                                 }
1778                         }
1779                 }
1780         }
1781
1782         private static function discoverServer(array $data, $default_generation = 0)
1783         {
1784                 if (empty($data['entry'])) {
1785                         return false;
1786                 }
1787
1788                 $success = false;
1789
1790                 foreach ($data['entry'] as $entry) {
1791                         $profile_url = '';
1792                         $profile_photo = '';
1793                         $connect_url = '';
1794                         $name = '';
1795                         $network = '';
1796                         $updated = NULL_DATE;
1797                         $location = '';
1798                         $about = '';
1799                         $keywords = '';
1800                         $gender = '';
1801                         $contact_type = -1;
1802                         $generation = $default_generation;
1803
1804                         if (!empty($entry['displayName'])) {
1805                                 $name = $entry['displayName'];
1806                         }
1807
1808                         if (isset($entry['urls'])) {
1809                                 foreach ($entry['urls'] as $url) {
1810                                         if ($url['type'] == 'profile') {
1811                                                 $profile_url = $url['value'];
1812                                                 continue;
1813                                         }
1814                                         if ($url['type'] == 'webfinger') {
1815                                                 $connect_url = str_replace('acct:' , '', $url['value']);
1816                                                 continue;
1817                                         }
1818                                 }
1819                         }
1820
1821                         if (isset($entry['photos'])) {
1822                                 foreach ($entry['photos'] as $photo) {
1823                                         if ($photo['type'] == 'profile') {
1824                                                 $profile_photo = $photo['value'];
1825                                                 continue;
1826                                         }
1827                                 }
1828                         }
1829
1830                         if (isset($entry['updated'])) {
1831                                 $updated = date(DateTimeFormat::MYSQL, strtotime($entry['updated']));
1832                         }
1833
1834                         if (isset($entry['network'])) {
1835                                 $network = $entry['network'];
1836                         }
1837
1838                         if (isset($entry['currentLocation'])) {
1839                                 $location = $entry['currentLocation'];
1840                         }
1841
1842                         if (isset($entry['aboutMe'])) {
1843                                 $about = HTML::toBBCode($entry['aboutMe']);
1844                         }
1845
1846                         if (isset($entry['gender'])) {
1847                                 $gender = $entry['gender'];
1848                         }
1849
1850                         if (isset($entry['generation']) && ($entry['generation'] > 0)) {
1851                                 $generation = ++$entry['generation'];
1852                         }
1853
1854                         if (isset($entry['contactType']) && ($entry['contactType'] >= 0)) {
1855                                 $contact_type = $entry['contactType'];
1856                         }
1857
1858                         if (isset($entry['tags'])) {
1859                                 foreach ($entry['tags'] as $tag) {
1860                                         $keywords = implode(", ", $tag);
1861                                 }
1862                         }
1863
1864                         if ($generation > 0) {
1865                                 $success = true;
1866
1867                                 logger("Store profile ".$profile_url, LOGGER_DEBUG);
1868
1869                                 $gcontact = ["url" => $profile_url,
1870                                                 "name" => $name,
1871                                                 "network" => $network,
1872                                                 "photo" => $profile_photo,
1873                                                 "about" => $about,
1874                                                 "location" => $location,
1875                                                 "gender" => $gender,
1876                                                 "keywords" => $keywords,
1877                                                 "connect" => $connect_url,
1878                                                 "updated" => $updated,
1879                                                 "contact-type" => $contact_type,
1880                                                 "generation" => $generation];
1881
1882                                 try {
1883                                         $gcontact = GContact::sanitize($gcontact);
1884                                         GContact::update($gcontact);
1885                                 } catch (Exception $e) {
1886                                         logger($e->getMessage(), LOGGER_DEBUG);
1887                                 }
1888
1889                                 logger("Done for profile ".$profile_url, LOGGER_DEBUG);
1890                         }
1891                 }
1892                 return $success;
1893         }
1894
1895 }